Currently, one of the biggest issues with LLMs is the limited context size. However, query time grows quadratically with that size, effectively putting an upper limit on it.
https://memgpt.ai/ is a new approach that offers what is effectively an unlimited context size.
It's available on #github: https://github.com/cpacker/MemGPT
#AI #LLMs #ChatGPT #OpenSource
GitHub - cpacker/MemGPT: Teaching LLMs memory management for unbounded context 📚🦙
Teaching LLMs memory management for unbounded context 📚🦙 - GitHub - cpacker/MemGPT: Teaching LLMs memory management for unbounded context 📚🦙GitHub
MonkeyPanic! :veri_trek2:
in reply to Curtis "Ovid" Poe (he/him) • • •Curtis "Ovid" Poe (he/him)
in reply to MonkeyPanic! :veri_trek2: • • •@MonkeyPanic I haven't had a chance to try it yet. I've been busy writing an immutable ORM for #Python to better learn the language.
Probably not the easiest project I could try, but I've done it in #Perl 😀